ABSTRACT
In order to test the two-variety model, it is necessary first to select binary morphological variables for which a Standard and a Dialect variant can be identified. The reduced form is in fact the most common variant in the Avion data for this variable, accounting for 476 out of 1150 tokens. Thelander investigated covaration of Standard and burtraskmal variants by means of a controlled experiment, involving fifty-six informants in fourteen two-hour group sessions, divided into two parts ‘A’ and ‘B’ by the introduction of a stranger from northern Sweden after one hour. There are a number of important differences between the Avion and Burträsk data. Thelander claims that the broad D-score range in his Burträsk data is indicative of two macrovariables rather than one, and thus of three rather than two varieties. The five highest ranked variables all have D-scores above 50%, and are separated by only twenty-three percentage points.
